Exploring causes of variation

Mundinger (1979) describes the basic structure of vocalizations as those fundamental, underlying species– typical features of a vocalization, uninfluenced by tradition (learning). Such characteristics may include aspects of the fundamental frequency, duration, tonal characteristics or pulsing of sound energy (Mundinger 1982b). Components of learned song may have a genetic basis in birds (for review see Slabbekoorn and Smith 2002a). The same may be true for bottlenose dolphins which show substantial population level genetic differentiation (reviewed in Chapter 5) and acoustic features may be honest signs of genetic diversity (Araya-Ajoy et al. 2009). The relationship between body length and call frequency transcends taxonomic divides (Badyaev and Leaf 1997, Matthews et al. 1999, Ding et al. 2001, Bertelli and I. 2002, May-Collado et al. 2007a). In general, larger

animals produce lower frequency calls. Several authors have suggested that body size may explain inter-specific differences in whistle frequency parameters (Wang et al. 1995a, Matthews et al. 1999). However the relationship is less strong when phylogenetic relationships are taken into account (May-Collado et al. 2007b) and recent evidence shows that killer whales, the largest delphinid, produce whistles with fundamental frequencies ranging up to 75 kHz (Samarra et al. 2010).Comparisons of Tursiops morphology from regions where the distributions of the two species (forms) overlap (SE Africa, Eastern Australia, E & S China Seas), have found no overlap in body lengths between the two species and only small overlap in the skull lengths (Hale et al. 2000, Kurihara and Oda 2009). However, there is plasticity in both these measures which, at least for T. truncatus, is correlated with variation in sea surface temperate (Ross and Cockcroft 1990, Hale et al. 2000). In both species (forms) there is a positive relationship between body length and skull length, although the relationship is stronger in T. truncatus. Sexual dimorphism in body length occurs in some populations of T. truncatus, but not T. aduncus, although this is not clearly apparent in skull measurements (Tolly et al. 1995). Body and skull lengths of inshore T. truncatus from the Gulf of Mexico (c.f. Florida, this study) are comparable with several geographically distinct populations of T. aduncus (Hale et al. 2000).

One might expect intra-specific variation to be constrained to occur within a species specific framework (Wilcznski and Ryan 1999). However, this need not be the case in a behaviourally plastic and cosmopolitan genus such as Tursiops spp., where the capacity for vocal production learning, might enable adaptation to local environmental conditions which transcends species boundaries. This ‘habitat adaptation hypothesis;’ has been widely applied in the study of birds (Morton 1975, Date and Lemon 1993), which like dolphins demonstrate vocal production learning. Here, divergent bird species living in the same habitat converge on calls with similar properties which transmit better within the habitat in which they are produced. Degradation is habitat dependent and has been described as the ‘sum of all the changes in the signal at distance X relative to the signal's structure at its origin or source’ (Morton 1986). It generally results in three structural changes in sound: reverberation, irregular amplitude fluctuations and frequency dependent attenuation (Naguib and Wiley 2001, Naguib 2003). Under this definition, a uniform decrease in the amplitude among all components in not viewed as degradation (Morton 1986). For signature whistles to be effective as contact calls, features which promote reliable, long range, signal transmission and minimise degradation within a particular habitat may be selected for.

Tests for acoustic adaptation have been carried out in many taxon and habitats, with varying support (Badyaev and Leaf 1997, Daniel and Blumstein 1998, Couldridge and van Staaden 2004, Boncoraglio and Saino 2007). There has been little dedicated research to test the acoustic adaptation hypothesis in dolphin vocalisations. However, habitat type can influence the structural properties of bottlenose dolphin signature whistles (Buckstaff 2004) and both signal frequency characteristics and levels of whistle modulation vary with acoustic environment which may facilitate long range transmission (Morisaka et al. 2005a). Habitat variations including depth, substrate, seagrass cover and the occurrence of channels can influence whistle propagation (Quintana-Rizzo et

al. 2006) and may impact on signature whistle form. In general, the degree of transmission loss is dependent on whistle frequency, with lower frequency whistles travelling greater distance. In frequency modulated whistles, the lower frequency components will travel further than higher frequency elements (Janik 2000a). However, in seagrass habitats low frequency whistles attenuate more and thus propagate over shorter distances compared to the same whistles used in a sandy mud habitat (Quintana-Rizzo et al. 2006). In acoustically noisy habitats alternative mechanisms for enhancing the detectability of sounds include; increasing the call rate, increasing signal intensity, increasing signal duration and shifting frequency outside the noise band (Tyack 2008).

Our understanding of the influence of depth on whistle production is limited to a handful of studies, and the outcomes are inconsistent. For Beluga whales, (Delphinapterus leucas), there is evidence that whistle amplitude decreases with depth whilst the frequency of the highest intensity peak increases with depth (Ridgway et al. 2001). Reduction in amplitude may reflect the increasing density of air in the nasal region where sound is generated (Ridgway et al. 2001). However, more recent work using digital archive tags (DTAGS), (Acevedo- Gutierrez 2002) suggests that behavioural and not biophysical explanations may underlie the change in Beluga whale call frequency identified by Ridgway et al. (2001). In short-finned pilot whales (Globicephala macrorhynchus) the frequency of social sounds produced at depth vary little from those produced near the surface. However, the acoustic energy output of calls was an order of magnitude less than that produced at the surface and the duration of calls was shorter (Acevedo-Gutierrez 2002). Both these studies are based on species with ecological and cranial characteristics different to those of bottlenose dolphins. The only study which has investigated the characteristics of signature whistles produced at varying depths found no consistent effect on maximum frequency, duration or whistle rate. However, minimum frequency of signature whistles was significantly lower in deeper water habitats, expanding the bandwidth of whistles (Buckstaff 2004).

There is strong evidence that signature whistle development involves learning (Caldwell and Caldwell 1979, Miksis et al. 2002, Fripp et al. 2005) although transmission might not be vertical (mother-offspring) but horizontal (through associates) (Sayigh et al. 1995, Fripp et al. 2005). Therefore, vocal production learning of signature whistle form or features can be viewed as cultural transmission (Slater 1986), potentially creating greater similarity within populations than between (Fripp et al. 2005). Changes in these features over time (cultural drift), whilst other variables remain unchanged is further evidence for learned transmission in call type (Deecke et al. 2000, Noad et al. 2000) and may enhance differences between populations. Directional changes, such as call convergence (Tyack 2008) between alliance members can reduce diversity in whistle types (Smolker and Pepper 1999, Watwood et al. 2004).

Broad scale studies of the evolution of whistle complexity have shown more complex whistle structure to correlate positively with social structure (May-Collado et al. 2007a) .Bottlenose dolphins live in fission-fusion societies and both the number of associates (group) and conspecifics sharing the same acoustic environment

(population) may enhance selection for more diverse and complex whistle types. Furthermore, stereotypy might be expected to increase in larger populations, if recognition depends on reliable transmission of subtle differences in one or two acoustic features. Inter-individual encounter rates pose a challenge to identity encoding and individual recognition, as well as to signature whistle transmission. Whistling rates vary with behavioural context and between geographic regions (Jones and Sayigh 2002, Cook et al. 2004, Quick and Janik 2008). The relationship between whistle rates and group size is not necessarily linear and individual whistling rates may decrease in large group sizes to avoid masking by conspecifics (Quick and Janik 2008).

7.1.2. Objectives

The suite of potential influences on the characteristics of signature whistle types is clearly broad, diverse and potentially inter-linked. Unravelling these relationships is complicated by the difficulty in getting equivalent, biologically meaningful data for a wide range of populations. Here I examine the relationship between potential causes of variation on multiple aspects of bottlenose dolphin signature whistle types, to investigate the underlying causes of population and species variation in these calls. In particular, the objectives of this chapter are to:

1) Determine explanatory variables which may influence variation in SWTs and report on those where equivalent data are available for all populations (supplied in Appendix 14).

2) Use principal components analysis (PCA) as a data reduction technique to reduce the number of putative explanatory variables to those which are uncorrelated and explain the most variation in key characteristics of signature whistle types.

3) Highlight the explanatory variables which are likely to promote variation in signature whistle types using partial Mantel test correlations.

7.2. Methods

Eleven explanatory variables were investigated for their potential influence on signature whistle characteristics (Appendix 14). These factors can be broadly divided into those relating to species (inter- intra-specific, genetic distance, body size), social dynamics (population size, average group size, con-specific jamming, geographic distance, residency), and habitat characteristics (depth, latitude).